Generates fitted values, including bootstrap confidence intervals, for in- and out-of-sample data from a fitted polywog model.
# S3 method for polywog
predict(object, newdata, type = c("link", "response"),
interval = FALSE, level = 0.95, bag = FALSE, na.action = na.pass, ...)
a fitted model of class "polywog"
, typically the output
of polywog
.
an optional data frame containing observations for which fitted values should be computed. If not specified, fitted values are generated for the data used to fit the model.
specifies whether the fitted values should be generated on the link scale (\(X \beta\)) or in terms of the expected value of the response variable. These only differ for binomial family models.
logical: whether to calculate bootstrap confidence intervals for each fitted value.
confidence level for the intervals.
logical: whether to use "bootstrap aggregation" to generate the
main fitted values (if FALSE
, they are calculated from the main model
fit).
a function specifying what to do with observations in
newdata
containing NA
s (default na.pass
). See
"Details".
other arguments, currently ignored.
If interval = TRUE
, a matrix containing each fitted value and
its confidence interval. Otherwise, a vector containing the fitted values.
For more user-friendly generation of fitted values, see
predVals
. To compute marginal effects, see
margEff.polywog
.